在Vagrant上使用docker图像构建Redis群集：无法ping机器

我试图在OSX上本地构建一个简单的redis集群，在virtualbox虚拟机中使用docker。具体来说，我想创建一个主redis节点（192.168.0.100），其中包含两个从节点（.101和.102）。这涉及到配置每台机器，并向每台从机发出命令，告诉它键值存储关系中的主机是谁

这非常有效，直到我决定在Vagrant文件中做一些编程工作，使配置更加灵活。结果是，我甚至无法ping.102之外的任何机器，更不用说使用redis cli命令（例如，

redis cli-h 192.168.0.100 INFO

来查看主设备是否正确设置了从设备）

我的目标/限制：

• 在virtualbox虚拟机中运行docker：我在阅读文档时看到了几种方法。我选择这样做是因为我不想引用大量的外部文件
• 配置后，在每个VM上运行一系列命令，以设置从机等
• 我不想只使用boot2docker，因为我还有很多其他的东西要添加。。。twemproxy、sentinel和我想至少稍微模拟一下生产环境

我在新文件中做错了什么

编辑：我做了更多的调查。我尝试过的一件事是销毁所有图像，然后运行旧的vagrant文件来构建机器，然后用新的vagrant文件重新加载它们。在这个场景中，机器正常工作，但每次我都会看到身份验证失败消息，如

redis1:Warning:authentication failure。在机器启动后重试…

。不过，我还是能把这台机器砰砰地一声关上

---

新文件（使用下面的配置文件）：

config.rb：

module RedisClusterConfig
    attr_reader :redisNodes 

    RedisNode = Struct.new("RedisNode", :name, :address, :port, :slaveOfAddress, :slaveOfPort)

    p = "6379" # Default redis port
    baseAddr = "192.168.0."

    # Define a triplet
    redis0 = RedisNode.new("redis0", baseAddr + "100", p, nil, nil)
    redis1 = RedisNode.new("redis1", baseAddr + "101", p, baseAddr + "100", p)
    redis2 = RedisNode.new("redis2", baseAddr + "102", p, baseAddr + "100", p)

    @redisNodes = [redis0, redis1, redis2]
end

---

旧文件：

# Make sure you have triggers plugin:
#     vagrant plugin install vagrant-triggers
#
# TODO:
# - We really would like to run some shell commands after everything is done, not after each machine is provisioned

Vagrant.configure(2) do |config|
    config.vm.provider "virtualbox"

    ## Redis Nodes ##

    ### Master Node###
    config.vm.define "redis0" do |a|
        a.vm.hostname = "redis0"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.100"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    ### Slave Nodes ###
    config.vm.define "redis1" do |a|
        a.vm.hostname = "redis1"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.101"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    config.vm.define "redis2" do |a|
        a.vm.hostname = "redis2"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.102"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    # After everything else has finished provisioning...
    config.trigger.after [:up, :resume, :reload] do
        # Configuring redis master/slave triplet
        system('echo Configuring redis master/slave triplet...')
        system('redis-cli -h 192.168.0.101 SLAVEOF 192.168.0.100 6379')
        system('redis-cli -h 192.168.0.102 SLAVEOF 192.168.0.100 6379')
    end

end

---

//，快一年了。你有没有找到解决这个问题的方法，或者你必须从头开始重新做呢？我放弃了，直接在谷歌计算引擎中构建虚拟机，实际的系统就在那里。我现在正在重建系统以使用Kubernetes（可能是通过托管版本，Google容器引擎），我认为这将使流浪部分变得更容易，因为Kubernetes修复了Docker的坏网络。//，我认为Swarm修复了Docker的坏网络？//，已经快一年了。你有没有找到解决这个问题的方法，或者你必须从头开始重新做呢？我放弃了，直接在谷歌计算引擎中构建虚拟机，实际的系统就在那里。我现在正在重建系统以使用Kubernetes（可能是通过托管版本，Google容器引擎），我认为这将使流浪部分更容易，因为Kubernetes修复了Docker的坏网络。//，我想Swarm修复了Docker的坏网络？

# Make sure you have triggers plugin:
#     vagrant plugin install vagrant-triggers
#
# TODO:
# - We really would like to run some shell commands after everything is done, not after each machine is provisioned

Vagrant.configure(2) do |config|
    config.vm.provider "virtualbox"

    ## Redis Nodes ##

    ### Master Node###
    config.vm.define "redis0" do |a|
        a.vm.hostname = "redis0"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.100"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    ### Slave Nodes ###
    config.vm.define "redis1" do |a|
        a.vm.hostname = "redis1"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.101"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    config.vm.define "redis2" do |a|
        a.vm.hostname = "redis2"
        a.vm.box = "ubuntu/trusty64"

        # Skip checking for an updated Vagrant box
        a.vm.box_check_update = false

        # Networking
        a.vm.network :private_network, ip: "192.168.0.102"
        a.ssh.forward_agent = true
        a.ssh.insert_key = false

        a.vm.provision "docker" do |d|
          d.images = ["redis"]
          d.run "redis",
            name: "redis",
            args: "-p 6379:6379"
        end
    end

    # After everything else has finished provisioning...
    config.trigger.after [:up, :resume, :reload] do
        # Configuring redis master/slave triplet
        system('echo Configuring redis master/slave triplet...')
        system('redis-cli -h 192.168.0.101 SLAVEOF 192.168.0.100 6379')
        system('redis-cli -h 192.168.0.102 SLAVEOF 192.168.0.100 6379')
    end

end